Image credit : TechCrunch
Vibe coding is making application development faster, but it is also creating a new form of shadow IT. Employees can now build apps with sensitive business data before security teams know those apps exist.
Superblocks has launched Superblocks 3.0, a private enterprise vibe-coding platform developed through a strategic partnership with Amazon Web Services. It allows business teams to build, import, and deploy AI-generated applications inside an organization’s existing AWS environment.
The platform can run within a company’s AWS virtual private cloud, keeping application code, data and AI inference inside its established security perimeter. Access, encryption, networking and audit controls remain governed through AWS services, while approved models run through Amazon Bedrock. Superblocks says customer prompts and data are not used to train those models.
Teams can import prototypes from tools including ChatGPT, Claude, Lovable, and Replit, then move them through an IT-approved path to production. When an application requires storage, Superblocks can create Aurora or S3 resources inside the customer’s VPC and manage migrations between development and production environments.
Security is built into that process. Before code reaches production, specialized agents and deterministic scanners check for weaknesses such as SQL injection, exposed secrets, permission bypasses and unsafe data flows. The platform also maintains a software bill of materials and continuously monitors deployed dependencies for newly disclosed vulnerabilities.
Superblocks 3.0 also introduces a Smart Router that divides application-building tasks across approved frontier and open-weight models. The company says this model routing can reduce inference costs by up to 30%, although the figure remains company-reported.
The launch reflects a wider shift toward governed enterprise AI development, where companies want faster application creation without losing control of data, permissions or production software. Superblocks’ challenge will be proving that business teams can build securely at scale without creating a new layer of difficult-to-maintain applications.